CN109033995A - Identify the method, apparatus and intelligence wearable device of user behavior - Google Patents
Identify the method, apparatus and intelligence wearable device of user behavior Download PDFInfo
- Publication number
- CN109033995A CN109033995A CN201810718089.3A CN201810718089A CN109033995A CN 109033995 A CN109033995 A CN 109033995A CN 201810718089 A CN201810718089 A CN 201810718089A CN 109033995 A CN109033995 A CN 109033995A
- Authority
- CN
- China
- Prior art keywords
- user behavior
- convolutional neural
- neural networks
- initial data
- networks model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The embodiment of the invention discloses a kind of method, apparatus for identifying user behavior and intelligent wearable devices, are related to terminal applies technical field, and main purpose is that solving the problem of to identify that user behavior error is larger does not adapt to compound action.Technical solution of the present invention includes: to obtain the corresponding initial data of user behavior；The initial data is inputted into convolutional neural networks model, and the corresponding user behavior classification of the initial data is identified according to the filter parameter in the convolutional neural networks model.
Description
Technical field
The present embodiments relate to terminal applies technical fields, more particularly to a kind of method for identifying user behavior, dress
It sets and intelligent wearable device.
Background technique
Along with the fast development of intelligent wearable device industry, the application field of intelligent wearable device is also increasingly wider
It is general.Such as: smartwatch can be with real-time monitoring user behavior, to be more applications other than providing temporal information
Data source is provided, more intelligent usage experience is brought to user.
Currently, being accelerated when intelligent wearable device identifies user behavior using original sensing data, including fusion
Degree is counted with the data of gyroscope and extracts some simple features, is classified by simple threshold value to these features, with reality
The identification of current family behavior state.
Inventor has found in the prior art in the specific implementation process, collected original when identifying user behavior
Sensing data is often excessively ideal, has usually manually divided, what cannot be varied with each individual judges, the knowledge of user behavior
Other result is affected by extraneous factor, therefore the error that there is identification user behavior is larger, does not adapt to compound action
Problem.In this way, cannot identify complicated user behavior since the accuracy to user behavior recognition is inadequate, reduce use
Family experience.
Summary of the invention
In view of this, a kind of method, apparatus for identifying user behavior provided in an embodiment of the present invention and intelligence wearable are set
Standby, main purpose is that solving the problem of to identify that user behavior error is larger does not adapt to compound action.
To solve the above-mentioned problems, the embodiment of the present invention mainly provides the following technical solutions:
In a first aspect, the embodiment of the invention provides a kind of methods for identifying user behavior, this method comprises:
Obtain the corresponding initial data of user behavior；
The initial data is inputted into convolutional neural networks model, and according to the filtering in the convolutional neural networks model
Device parameter identifies the corresponding user behavior classification of the initial data.
Optionally, before the initial data is inputted convolutional neural networks model, the method also includes:
Flag data based on user behavior classification is trained the convolutional neural networks model.
Optionally, the flag data based on user behavior classification is trained packet to the convolutional neural networks model
It includes:
The initial data of known users behavior classification is marked, the flag data is obtained；The known users row
It include at least two for classification；
Extract the characteristic in the flag data；
The characteristic is inputted the convolutional neural networks model to be trained, obtains convolutional neural networks model
Filter parameter, each layer in convolutional neural networks model represent a filter parameter, the filter parameter and user
There are mapping relations for behavior classification.
Optionally, the filter parameter according in the convolutional neural networks model identifies that the initial data is corresponding
User behavior classification include:
Based in the convolutional neural networks model, the corresponding every layer of filter parameter of every kind of user behavior classification is to described
Identification is compared in initial data；
Determine similarity of the initial data respectively with every kind of user behavior classification；
Using the maximum user behavior classification of similarity as the corresponding user behavior classification of the initial data.
Optionally, include: by initial data input convolutional neural networks model
Input window based on the convolutional neural networks model executes format specification processing to the initial data；
The characteristic of initial data after extraction process, and the characteristic is input to by input window described
In convolutional neural networks model.
Optionally, the method also includes:
After the completion of to the convolutional neural networks model training, the test data based on user behavior classification is to the volume
Product neural network model is tested, wherein the test data is to mark the data of user behavior classification.
Optionally, the initial data is obtained from least one device in following device, and described device includes: three axis
Accelerometer, gyroscope, thermometer and Cardiotachometer.
Second aspect, the embodiment of the present invention also provide a kind of device for identifying user behavior, which includes:
Acquiring unit, for obtaining the corresponding initial data of user behavior；
Input unit, the initial data for obtaining the acquiring unit input convolutional neural networks model；
Recognition unit, for identifying the initial data pair according to the filter parameter in the convolutional neural networks model
The user behavior classification answered.
Optionally, described device further include:
Training unit is used for before the initial data is inputted convolutional neural networks model by the input unit, base
The convolutional neural networks model is trained in the flag data of user behavior classification.
Optionally, the training unit includes:
Mark module is marked for the initial data to known users behavior classification, obtains the flag data；Institute
Stating known users behavior classification includes at least two；
Extraction module, for extracting the characteristic in the flag data that first mark module obtains；
Training module, the characteristic for obtaining the extraction module input the convolutional neural networks model
It is trained, obtains the filter parameter of convolutional neural networks model, each layer in convolutional neural networks model represents one
Filter parameter, there are mapping relations with user behavior classification for the filter parameter.
Optionally, the recognition unit includes:
Comparison module, for based in the convolutional neural networks model, corresponding every layer of every kind of user behavior classification to be filtered
Identification is compared to the initial data in wave device parameter；
Determining module, for determining similarity of the initial data respectively with every kind of user behavior classification, by similarity
Maximum user behavior classification is as the corresponding user behavior classification of the initial data.
Optionally, the input unit includes:
Processing module, the input window based on the convolutional neural networks model execute format specification to the initial data
Processing；
Input module, the characteristic of the initial data after extraction process, and the characteristic is passed through into input window
It is input in the convolutional neural networks model.
Optionally, described device further include:
Test cell, for after the completion of training unit training, the test data based on user behavior classification to be to institute
It states convolutional neural networks model to be tested, wherein the test data is to mark the data of user behavior classification.
Optionally, the initial data is obtained from least one device in following device, and described device includes: three axis
Accelerometer, gyroscope, thermometer and Cardiotachometer.
The third aspect, the embodiment of the present invention also provide a kind of intelligent wearable device, and the intelligence wearable device includes:
At least one processor；
And at least one processor, the bus being connected to the processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor is used to call the program instruction in the memory, to execute described in any one of first aspect
The method for identifying user behavior.
Fourth aspect, the embodiment of the present invention also provide a kind of non-transient computer readable storage medium, the non-transient meter
Calculation machine readable storage medium storing program for executing stores computer instruction, and the computer instruction makes the computer execute any one of first aspect
The method of the identification user behavior.
By above-mentioned technical proposal, technical solution provided in an embodiment of the present invention is at least had the advantage that
The method, apparatus of identification user behavior provided in an embodiment of the present invention and intelligent wearable device, are used firstly, obtaining
The initial data is inputted convolutional neural networks model by the corresponding initial data of family behavior, and according to the convolutional Neural net
Filter parameter in network model identifies the corresponding user behavior classification of the initial data.With in the prior art generally use
Using original sensing data, some simple features are extracted, are classified by simple threshold value to these features, with reality
The identification of current family behavior state is compared, and present invention optimizes the method for user behavior recognition, is increased using convolutional Neural net
Network model identifies user behavior, reduces the error of identification user behavior, and can identify more complicated user behavior.
Above description is only the general introduction of technical solution of the embodiment of the present invention, in order to better understand the embodiment of the present invention
Technological means, and can be implemented in accordance with the contents of the specification, and in order to allow above and other mesh of the embodiment of the present invention
, feature and advantage can be more clearly understood, the special specific embodiment for lifting the embodiment of the present invention below.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention
The limitation of embodiment.And throughout the drawings, the same reference numbers will be used to refer to the same parts.In the accompanying drawings:
Fig. 1 shows a kind of flow chart of the method for identification user behavior that embodiment disclosed by the invention provides；
Fig. 2 shows the flow charts for another method for identifying user behavior that embodiment disclosed by the invention provides；
Fig. 3 shows a kind of composition block diagram of the device for identification user behavior that embodiment disclosed by the invention provides；
Fig. 4 shows the composition block diagram of the device for another identification user behavior that embodiment disclosed by the invention provides；
Fig. 5 shows a kind of frame of the intelligent wearable device for identification user behavior that embodiment disclosed by the invention provides
Schematic diagram.
Specific embodiment
In the prior art, when to user behavior recognition, original sensing data collected is often excessively ideal, uses
The recognition result hand extraneous factor of family behavior is affected, and the accuracy that may result in user behavior recognition is inadequate, no
It can identify complicated user behavior, reduce user experience.
To solve the above-mentioned problems, embodiment disclosed by the invention, using convolutional neural networks model to user behavior into
Row identification, reduces the error of identification user behavior, and can identify that more complicated user behavior embodiment disclosed by the invention mentions
For a kind of method for identifying user behavior, as shown in Figure 1, which comprises
101, the corresponding initial data of user behavior is obtained.
In practical applications, intelligent wearable device described in embodiment disclosed by the invention may include but be not limited to
The following contents, such as: smartwatch, Intelligent bracelet, intelligent shoe, intelligent head hoop, smart phone etc., specifically without limitation,
Embodiment disclosed by the invention to concrete type, model of intelligent wearable device etc. without limitation.For the ease of subsequent implementation
The description of example, subsequent embodiment are illustrated by taking smartwatch as an example, but it should be clear that this kind illustrates that mode is not anticipated
The intelligent wearable device is being limited as smartwatch.
In embodiment disclosed by the invention, in smartwatch in use, the sensor in smartwatch can obtain user's row
For corresponding initial data, for embodying user behavior, initial data may include but is not limited to following the initial data
Content, such as: initial data is obtained by three axis accelerometer, gyroscope, thermometer and Cardiotachometer etc., specifically
Without limitation.The time interval of the sensor outflow initial data can be various ways, mode one: spread out of original number in real time
According to；Mode two: separated in time spreads out of an initial data, when user will be in a kind of behavior state for a long time, for example, with
Family taken exercise by running, swimming exercise when, can choose every ten minutes initial data of outflow or per every other hour
An initial data is spread out of, specific interval time can be arranged by user oneself.To facilitate the behavior to user to identify.
102, the initial data is inputted into convolutional neural networks model, and according in the convolutional neural networks model
Filter parameter identifies the corresponding user behavior classification of the initial data.
In embodiment disclosed by the invention, in order to reduce the error of identification user behavior, and more complicated use can be identified
Family behavior, for example, it is upper downstairs, stand, drive, walking, running, butterfly stroke, breaststroke etc., there may be classification for those user behaviors
More, still, the little situation of difference between every kind of classification in embodiment disclosed by the invention, will acquire user behavior pair
The initial data answered is input in convolutional neural networks model and identifies to user behavior.
Each layer of convolutional neural networks model is all the filter of a sliding, can then be deposited in convolutional neural networks model
In multiple filters, by the difference of the parameter of filter, each layer can filter out different features, so in identification user
It is handed on, is determined according to the result of convolutional neural networks model output final from level to level by comparing different features when behavior
The corresponding user behavior classification of initial data.The output the result is that vector form, for example, the user behavior to be identified is
It is upper go downstairs, stand, sitting down, walking, breaststroke, 7 classes such as butterfly stroke, add other classes, then the output of convolutional neural networks model
Result vector be then 1 probability matrix for multiplying 8, this 8 probability and be 1, maximum probability user behavior is then this section of original number
According to user behavior.
It should be noted that filter parameter described in the embodiment of the present invention its be substantially hidden layer weight matrix, each filter
The weighted of wave device parameter, when identifying user behavior, according to the corresponding weight calculation probability of the parameter of different filters
Size.
The method for the identification user behavior that embodiment disclosed by the invention provides, firstly, obtaining the corresponding original of user behavior
The initial data is inputted convolutional neural networks model by beginning data, and according to the filtering in the convolutional neural networks model
Device parameter identifies the corresponding user behavior classification of the initial data.With generallyd use in the prior art using original sensing
Device data extract some simple features, are classified by simple threshold value to these features, to realize user behavior state
Identification compare, present invention optimizes the method for user behavior recognition, increase using convolutional neural networks model to user's row
To be identified, the error of identification user behavior is reduced, and can identify more complicated user behavior.
As the refinement and extension to above-described embodiment, in embodiment disclosed by the invention, to identification user behavior
Method can advanced optimize, and before being identification user behavior, be first trained to convolutional neural networks model, improved knowledge
The accuracy of other user behavior.In order to realize the above functions, the embodiment of the present invention also provides a kind of identification user behavior side of being put into
Method, as shown in Figure 2, which comprises
201, the flag data based on user behavior classification is trained the convolutional neural networks model.
In embodiment disclosed by the invention, before identifying user behavior by convolutional neural networks model, need to volume
Product neural network model is trained, and to obtain the filter parameter of convolutional neural networks model, passes through convolutional neural networks mould
The available different feature of difference of filter parameter in type.
Before convolutional neural networks are trained, need to obtain the initial data of a large amount of known users behavior classification,
The initial data of acquisition may include but be not limited to the following contents, such as: pass through three axis accelerometer, gyroscope, thermometer
And the data that Cardiotachometer etc. obtains, specifically without limitation.And these data are divided into training set and test set, institute
Stating training set is for training convolutional neural networks model, and the test set is for testing convolutional neural networks
's.The initial data of the behavior classification of acquisition may include but be not limited to the following contents, such as: above going downstairs, stand, sitting down,
It drives, walk, running, breaststroke, the corresponding initial data of butterfly stroke etc., trained user behavior classification is namely using intelligently
The user behavior classification that can be identified when wrist-watch.
When to convolutional neural networks model training, following manner may be employed without limitation of: corresponding to user behavior
Initial data with statistical method extract uniform format characteristic, the statistical method can be average,
Variance, spectrum energy etc. are able to reflect the characteristic of user behavior, specifically without limitation.In training convolutional nerve net
When network model, select the training distinguished every a kind of user behavior, training needed after completing to this kind of user behavior into
Line flag, illustratively, it is to walk that first trained, then corresponding initial data of walking is needed to be input to convolutional Neural net
The input window of network model, the input window can execute format specification processing to the initial data, and execute original number
According to characteristic extract, characteristic is input in convolutional neural networks model and is trained filter parameter, is obtained
The filter parameter with walk there are mapping relations, before executing convolutional neural networks model, first to user behavior classification
It is marked, so that in the initial data of training label reference can be provided to the matching of user behavior classification.This is trained
Result queue is that the representative of number 1 is walked, and is successively trained label to other user behaviors, number 2 is represented standing etc.,
Every kind of user behavior is numbered, final user behavior classification is obtained.It should be noted that convolutional neural networks mould
When type is trained, the characteristic of input is more, and corresponding training process is more accurate.
202, after the completion of to the convolutional neural networks model training, the test data based on user behavior classification is to institute
It states convolutional neural networks model to be tested, wherein the test data is to mark the data of user behavior classification.
In embodiment disclosed by the invention, after the completion of convolutional neural networks model training, to convolutional Neural pessimistic concurrency control
It is tested, to test the accuracy of convolutional neural networks model identification user behavior, the test data is answered for known users
For the data of the test set of classification.Test convolutional neural networks model when, actually identify user behavior process, will
In the initial data input convolutional neural networks model of known users behavior classification, pass through the input window of convolutional neural networks model
Mouthful to initial data carry out standardization processing after, the corresponding every layer of filter parameter of user behavior classification to the initial data into
Row matching identification is obtained using the maximum user behavior classification of similarity as the corresponding user behavior classification of the initial data
Output is as a result, judge whether the behavior classification of the result exported and label is consistent, determines the training of convolutional neural networks model
Accuracy.
203, the corresponding initial data of user behavior is obtained (with step 101).
Explanation in relation to step 203, please refers to the detailed description of step 101, and the embodiment of the present invention is no longer gone to live in the household of one's in-laws on getting married herein
It states.
204, the initial data is inputted into convolutional neural networks model, and according in the convolutional neural networks model
Filter parameter identifies the corresponding user behavior classification of the initial data.
In embodiment disclosed by the invention, after convolutional neural networks model obtains user behavior initial data, convolution
The Unified Form that original data processing is format specification need to just can be carried out identification by neural network model.By the original after extraction process
The characteristic of beginning data is input in the convolutional neural networks model by input window, and convolutional neural networks model is deposited
Initial data character pair data, the lattice are obtained for executing format specification processing to the initial data in input window
Formula specification handles are to extract characteristic by statistical method, and illustratively, the statistical method, which can be, to be asked
Average value, variance, spectrum energy etc. are able to reflect the characteristic of user behavior, specifically without limitation.
The classification of the user behavior of convolutional neural networks model identification is trained when being with training convolutional neural networks model
User behavior classification be consistent, illustratively, in training convolutional neural networks, trained user behavior classification is upper and lower
Building stands, sits down, running, breaststroke, 9 classes such as butterfly stroke, adds other classes, and distinguish label to user behavior, exemplary
, number of standing is 1, upper number downstairs is 2, breaststroke 3, butterfly stroke 4, other classes are 10 etc., then convolutional Neural net at this time
Network model output result vector be then 1 probability matrix for multiplying 10, it should be noted that this 10 probability and be 1.According to instruction
Practice the corresponding filter parameter of every kind of user behavior obtained by convolutional neural networks model, compare identification with initial data,
Phase knowledge and magnanimity of the initial data respectively with every kind of user behavior classification are determined, using the maximum user behavior classification of phase knowledge and magnanimity as described in
The corresponding user behavior of initial data.Illustratively, the user behavior of sensor outflow is got in convolutional neural networks model
After corresponding initial data, the input window of convolutional neural networks model will execute format specification processing to initial data, and
Filter parameter corresponding to treated initial data every kind of user behavior classification resulting to training is compared into identification,
The similarity of initial data Yu every kind of user behavior is obtained after filtering from level to level, it should be noted that initial data and every kind
The similarity of user behavior and be 1.For example, when the similarity of the corresponding initial data of user behavior and standing is 0.1, with it is upper
Similarity downstairs is 0.05, and the similarity with breaststroke is 0.5, and the similarity with butterfly stroke is 0.2 etc., then other six classes is similar
Degree and be 0.15, then it be first position number is 0.1, second position that the output 1 of convolutional neural networks model, which multiplies 10 probability matrix,
Being set to 0.05, third position is the probability and be 1 that 0.5, the 4th position is 0.2 etc. ten position, it is possible thereby to determine former
The similarity highest of beginning data and breaststroke, then using breaststroke as the user behavior classification of initial data.
To sum up, by obtaining filter parameter to convolutional neural networks model training, and to trained convolutional neural networks
It is tested, obtains classification results；The resulting filter parameter of training based on convolutional neural networks model identifies user behavior,
The accuracy rate for improving identification user behavior, identifies increasingly complex user behavior.
Further, as the realization to method shown in above-mentioned Fig. 1 and Fig. 2, another embodiment of the embodiment of the present invention is also mentioned
A kind of device for identifying user behavior is supplied.The Installation practice is corresponding with preceding method embodiment, is easy to read, the present apparatus
Embodiment no longer repeats the detail content in preceding method embodiment one by one, it should be understood that the dress in the present embodiment
The full content realized in preceding method embodiment can be corresponded to by setting.
Embodiment disclosed by the invention provides a kind of device for identifying user behavior, as shown in Figure 3, comprising:
Acquiring unit 31, for obtaining the corresponding initial data of user behavior；
Input unit 32, the initial data for obtaining the acquiring unit input convolutional neural networks model；
Recognition unit 33, for identifying the initial data according to the filter parameter in the convolutional neural networks model
Corresponding user behavior classification.
The device for the identification user behavior that embodiment disclosed by the invention provides, firstly, obtaining the corresponding original of user behavior
The initial data is inputted convolutional neural networks model by beginning data, and according to the filtering in the convolutional neural networks model
Device parameter identifies the corresponding user behavior classification of the initial data.With generallyd use in the prior art using original sensing
Device data extract some simple features, are classified by simple threshold value to these features, to realize user behavior state
Identification compare, present invention optimizes the method for user behavior recognition, increase using convolutional neural networks model to user's row
To be identified, the error of identification user behavior is reduced, and can identify more complicated user behavior.
Further, as shown in figure 4, training unit 34, for the initial data to be inputted in the input unit 32
Before convolutional neural networks model, the flag data based on user behavior classification instructs the convolutional neural networks model
Practice.
Further, as shown in figure 4, the training unit 34 includes:
Mark module 341 is marked for the initial data to known users behavior classification, obtains the reference numerals
According to；The known users behavior classification includes at least two；
Extraction module 342, for extracting the characteristic in the flag data that first mark module 341 obtains
According to；
Training module 343, the characteristic for obtaining the extraction module 342 input the convolutional Neural net
Network model is trained, and obtains the filter parameter of convolutional neural networks model, each layer of generation in convolutional neural networks model
One filter parameter of table, there are mapping relations with user behavior classification for the filter parameter.
Further, as shown in figure 4, the recognition unit 33 includes:
Comparison module 331, for based in the convolutional neural networks model, every kind of user behavior classification to be every layer corresponding
Identification is compared to the initial data in filter parameter；
Determining module 332 will be similar for determining similarity of the initial data respectively with every kind of user behavior classification
Maximum user behavior classification is spent as the corresponding user behavior classification of the initial data.
Further, as shown in figure 4, the input unit 32 includes:
Processing module 321, the input window based on the convolutional neural networks model execute format to the initial data
Specification handles；
Input module 322, the characteristic of the initial data after extracting the processing unit processes, and by the characteristic
It is input in the convolutional neural networks model according to by input window.
Further, as shown in figure 4, described device further include:
Test cell 35 is used for after the completion of the training unit 34 training, the test data based on user behavior classification
The convolutional neural networks model is tested, wherein the test data is to mark the data of user behavior classification.
Further, the initial data is obtained from least one device in following device, and described device includes: three
Axis accelerometer, gyroscope, thermometer and Cardiotachometer.
To sum up, by obtaining filter parameter to convolutional neural networks model training, and to trained convolutional neural networks
It is tested, obtains classification results；The resulting filter parameter of training based on convolutional neural networks model identifies user behavior,
The accuracy rate for improving identification user behavior, identifies increasingly complex user behavior.
The device for the identification user behavior introduced by the present embodiment is the identification that can be executed in the embodiment of the present invention
The device of the method for user behavior, so based on the method for identifying user behavior described in the embodiment of the present invention, this field
Those of skill in the art can understand the specific embodiment and its various change of the device of the identification user behavior of the present embodiment
Form, so how the device at this for the identification user behavior realizes a variety of identification user behaviors in the embodiment of the present invention
Method be no longer discussed in detail.As long as those skilled in the art implement the method for identifying user behavior in the embodiment of the present invention
Used device belongs to the range to be protected of the application.
The device of the identification user behavior includes processor and memory, and above-mentioned acquiring unit, input unit, identification are single
Member etc. stores in memory as program unit, executes above procedure unit stored in memory by processor Lai real
Now corresponding function.
Include kernel in processor, is gone in memory to transfer corresponding program unit by kernel.Kernel can be set one
Or more, compound action is not adapted to by adjusting kernel parameter to solve the problem of to identify that user behavior error is larger.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/
Or the forms such as Nonvolatile memory, if read-only memory (ROM) or flash memory (flash RAM), memory include that at least one is deposited
Store up chip.
The embodiment of the invention provides a kind of non-transient computer readable storage medium, the non-transient computer is readable to be deposited
Storage media stores computer instruction, and the computer instruction makes the computer execute identification user row described in above-described embodiment
For method.
The embodiment of the invention provides a kind of intelligent wearable devices, as shown in Figure 5, comprising: at least one processor
(processor)51；And at least one processor (memory) 52, the bus 53 being connect with the processor 51；Wherein,
The processor 51, memory 52 complete mutual communication by the bus 53；
The processor 51 is used to call the program instruction in the memory 52, to execute following steps:
Obtain the corresponding initial data of user behavior；
The initial data is inputted into convolutional neural networks model, and according to the filtering in the convolutional neural networks model
Device parameter identifies the corresponding user behavior classification of the initial data.
Optionally, before the initial data is inputted convolutional neural networks model, the method also includes:
Flag data based on user behavior classification is trained the convolutional neural networks model.
Optionally, the flag data based on user behavior classification is trained packet to the convolutional neural networks model
It includes:
The initial data of known users behavior classification is marked, the flag data is obtained；The known users row
It include at least two for classification；
Extract the characteristic in the flag data；
The characteristic is inputted the convolutional neural networks model to be trained, obtains convolutional neural networks model
Filter parameter, each layer in convolutional neural networks model represent a filter parameter, the filter parameter and user
There are mapping relations for behavior classification.
Optionally, the filter parameter according in the convolutional neural networks model identifies that the initial data is corresponding
User behavior classification include:
Based in the convolutional neural networks model, the corresponding every layer of filter parameter of every kind of user behavior classification is to described
Identification is compared in initial data；
Determine similarity of the initial data respectively with every kind of user behavior classification；
Using the maximum user behavior classification of similarity as the corresponding user behavior classification of the initial data.
Optionally, include: by initial data input convolutional neural networks model
Input window based on the convolutional neural networks model executes format specification processing to the initial data；
The characteristic of initial data after extraction process, and the characteristic is input to by input window described
In convolutional neural networks model.
Optionally, the method also includes:
After the completion of to the convolutional neural networks model training, the test data based on user behavior classification is to the volume
Product neural network model is tested, wherein the test data is to mark the data of user behavior classification.
Optionally, the initial data is obtained from least one device in following device, and described device includes: three axis
Accelerometer, gyroscope, thermometer and Cardiotachometer.
Present invention also provides a kind of computer program products, when executing on data processing equipment, are adapted for carrying out just
The program code of beginningization there are as below methods step:
Obtain the corresponding initial data of user behavior；
The initial data is inputted into convolutional neural networks model, and according to the filtering in the convolutional neural networks model
Device parameter identifies the corresponding user behavior classification of the initial data.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/
Or the forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable Jie
The example of matter.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including element
There is also other identical elements in process, method, commodity or equipment.
It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The above is only embodiments herein, are not intended to limit this application.To those skilled in the art,
Various changes and changes are possible in this application.It is all within the spirit and principles of the present application made by any modification, equivalent replacement,
Improve etc., it should be included within the scope of the claims of this application.
Claims (10)
1. a kind of method for identifying user behavior characterized by comprising
Obtain the corresponding initial data of user behavior；
The initial data is inputted into convolutional neural networks model, and is joined according to the filter in the convolutional neural networks model
Number identifies the corresponding user behavior classification of the initial data.
2. the method according to claim 1, wherein the initial data is inputted convolutional neural networks model
Before, the method also includes:
Flag data based on user behavior classification is trained the convolutional neural networks model.
3. according to the method described in claim 2, it is characterized in that, the flag data based on user behavior classification is to described
Convolutional neural networks model, which is trained, includes:
The initial data of known users behavior classification is marked, the flag data is obtained；The known users behavior class
It Bao Han at least two；
Extract the characteristic in the flag data；
The characteristic is inputted the convolutional neural networks model to be trained, obtains the filtering of convolutional neural networks model
Device parameter, each layer in convolutional neural networks model represent a filter parameter, the filter parameter and user behavior
There are mapping relations for classification.
4. according to the method described in claim 3, it is characterized in that, the filtering according in the convolutional neural networks model
Device parameter identifies that the corresponding user behavior classification of the initial data includes:
Based in the convolutional neural networks model, the corresponding every layer of filter parameter of every kind of user behavior classification is to described original
Identification is compared in data；
Determine similarity of the initial data respectively with every kind of user behavior classification；
Using the maximum user behavior classification of similarity as the corresponding user behavior classification of the initial data.
5. according to the method described in claim 3, it is characterized in that, described input convolutional neural networks mould for the initial data
Type includes:
Input window based on the convolutional neural networks model executes format specification processing to the initial data；
The characteristic of initial data after extraction process, and the characteristic is input to the convolution by input window
In neural network model.
6. according to the method described in claim 3, it is characterized in that, the method also includes:
After the completion of to the convolutional neural networks model training, the test data based on user behavior classification is to the convolution mind
It is tested through network model, wherein the test data is to mark the data of user behavior classification.
7. method according to claim 1 to 6, which is characterized in that the initial data is from following device
It is obtained at least one device, described device includes: three axis accelerometer, gyroscope, thermometer and Cardiotachometer.
8. a kind of device for identifying user behavior characterized by comprising
Acquiring unit, for obtaining the corresponding initial data of user behavior；
Input unit, the initial data for obtaining the acquiring unit input convolutional neural networks model；
Recognition unit, for identifying that the initial data is corresponding according to the filter parameter in the convolutional neural networks model
User behavior classification.
9. a kind of intelligence wearable device characterized by comprising
At least one processor；
And at least one processor, the bus being connected to the processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor is used to call the program instruction in the memory, any into claim 7 with perform claim requirement 1
The method of identification user behavior described in.
10. a kind of non-transient computer readable storage medium, which is characterized in that the non-transient computer readable storage medium is deposited
Store up computer instruction, the computer instruction requires the computer perform claim 1 to described in any one of claim 7
The method for identifying user behavior.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810718089.3A CN109033995A (en) | 2018-06-29 | 2018-06-29 | Identify the method, apparatus and intelligence wearable device of user behavior |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810718089.3A CN109033995A (en) | 2018-06-29 | 2018-06-29 | Identify the method, apparatus and intelligence wearable device of user behavior |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109033995A true CN109033995A (en) | 2018-12-18 |
Family
ID=65522346
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810718089.3A Pending CN109033995A (en) | 2018-06-29 | 2018-06-29 | Identify the method, apparatus and intelligence wearable device of user behavior |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109033995A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961090A (en) * | 2019-02-28 | 2019-07-02 | 广州杰赛科技股份有限公司 | A kind of behavior classification method and system based on intelligent wearable device |
| CN110010224A (en) * | 2019-03-01 | 2019-07-12 | 出门问问信息科技有限公司 | User movement data processing method, device, wearable device and storage medium |
| CN111382403A (en) * | 2020-03-17 | 2020-07-07 | 同盾控股有限公司 | Training method, device, equipment and storage medium of user behavior recognition model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846729A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of fall detection method and system based on convolutional neural networks |
| CN107153871A (en) * | 2017-05-09 | 2017-09-12 | 浙江农林大学 | Fall detection method based on convolutional neural networks and mobile phone sensor data |
| CN107609501A (en) * | 2017-09-05 | 2018-01-19 | 东软集团股份有限公司 | The close action identification method of human body and device, storage medium, electronic equipment |
| CN108062170A (en) * | 2017-12-15 | 2018-05-22 | 南京师范大学 | Multi-class human posture recognition method based on convolutional neural networks and intelligent terminal |
-
2018
- 2018-06-29 CN CN201810718089.3A patent/CN109033995A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846729A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of fall detection method and system based on convolutional neural networks |
| CN107153871A (en) * | 2017-05-09 | 2017-09-12 | 浙江农林大学 | Fall detection method based on convolutional neural networks and mobile phone sensor data |
| CN107609501A (en) * | 2017-09-05 | 2018-01-19 | 东软集团股份有限公司 | The close action identification method of human body and device, storage medium, electronic equipment |
| CN108062170A (en) * | 2017-12-15 | 2018-05-22 | 南京师范大学 | Multi-class human posture recognition method based on convolutional neural networks and intelligent terminal |
Non-Patent Citations (1)
| Title |
|---|
| 石代伟等: "结合手机传感器和卷积神经网络的人体行为识别", 《电子技术与软件工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961090A (en) * | 2019-02-28 | 2019-07-02 | 广州杰赛科技股份有限公司 | A kind of behavior classification method and system based on intelligent wearable device |
| CN110010224A (en) * | 2019-03-01 | 2019-07-12 | 出门问问信息科技有限公司 | User movement data processing method, device, wearable device and storage medium |
| CN111382403A (en) * | 2020-03-17 | 2020-07-07 | 同盾控股有限公司 | Training method, device, equipment and storage medium of user behavior recognition model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110348580A (en) | Construct the method, apparatus and prediction technique, device of GBDT model | |
| CN107358247B (en) | Method and device for determining lost user | |
| CN109658455A (en) | Image processing method and processing equipment | |
| CN108334647A (en) | Data processing method, device, equipment and the server of Insurance Fraud identification | |
| CN108563548A (en) | Method for detecting abnormality and device | |
| CN110070101A (en) | Floristic recognition methods and device, storage medium, computer equipment | |
| CN109145766A (en) | Model training method, device, recognition methods, electronic equipment and storage medium | |
| CN109858948A (en) | The determination method and apparatus of power customer marketing strategy | |
| CN109033995A (en) | Identify the method, apparatus and intelligence wearable device of user behavior | |
| CN107590460B (en) | Face classification method, apparatus and intelligent terminal | |
| CN109598234A (en) | Critical point detection method and apparatus | |
| CN109359515A (en) | A kind of method and device that the attributive character for target object is identified | |
| CN109740113A (en) | Hyper parameter threshold range determines method, apparatus, storage medium and electronic equipment | |
| CN109447156A (en) | Method and apparatus for generating model | |
| CN104391879B (en) | The method and device of hierarchical clustering | |
| WO2020228283A1 (en) | Feature extraction method and apparatus, and computer readable storage medium | |
| CN108937867A (en) | A kind of dormant monitoring method and device | |
| CN110222728A (en) | The training method of article discrimination model, system and article discrimination method, equipment | |
| CN108509833A (en) | A kind of face identification method, device and equipment based on structured analysis dictionary | |
| CN108989581A (en) | A kind of consumer's risk recognition methods, apparatus and system | |
| CN110457677A (en) | Entity-relationship recognition method and device, storage medium, computer equipment | |
| CN109740567A (en) | Key point location model training method, localization method, device and equipment | |
| CN109345285A (en) | A kind of movable put-on method, device and equipment | |
| CN109614414A (en) | A kind of determination method and device of user information | |
| CN108717520A (en) | A kind of pedestrian recognition methods and device again |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |